Dear AFNI community,
I am having trouble choosing the predictors for a voxelwise multiple linear regression analysis (3dRegAna) where the dependent variable is a single regression coefficient. I have patient and control groups that do not differ significantly in terms of age or gender but do differ significantly in terms of education. The causal direction of the relationship between group and education is unclear. I would like to maximize my ability to detect group differences. Do I include education in the model? Should predictors that do not account for a significant amount of the variance be removed from the model?